EMD ultrasonic signal adaptive denoising method based on multi-index evaluation
By employing an EMD decomposition method based on multi-index evaluation, feature extraction and adaptive denoising of ultrasonic signals are performed, solving the problem of denoising and feature preservation in existing technologies and achieving efficient signal purification and feature extraction in lithium-ion battery detection.
Patent Information
- Application Number
- CN202511402227.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-09-28
- Publication Date
- 2025-11-28
AI Technical Summary
Existing ultrasonic signal denoising methods for lithium-ion battery testing suffer from problems such as strong parameter dependence, poor adaptability to non-stationary signals, and difficulty in balancing denoising and structural feature preservation, resulting in difficulty in ensuring detection accuracy and stability.
An EMD decomposition method based on multi-index evaluation is adopted to perform empirical mode decomposition on ultrasound signals, extracting features such as energy ratio, correlation coefficient, Shannon entropy, spectral entropy, and reconstruction error. The IMF components are scored by clustering algorithm and weight configuration, and the components with scores higher than the threshold are retained for signal reconstruction.
It effectively suppresses high-frequency noise and low-frequency drift, preserves key feature information, improves signal quality, and enhances the accuracy and stability of subsequent diagnostic judgments.
Smart Images
Figure CN121027308A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of ultrasonic nondestructive testing and signal processing technology, and in particular to an adaptive denoising method for EMD decomposition ultrasonic signals based on multi-index evaluation. Background Technology
[0002] Ultrasonic nondestructive testing is a detection technique that uses the properties of high-frequency sound waves to propagate, reflect, and scatter in materials to detect internal defects or performance changes in a target structure.
[0003] During ultrasonic testing, the received signal is often affected by multiple sources of interference, resulting in a large amount of clutter and noise in the original waveform, which can obscure the key echo structure or cause feature extraction errors.
[0004] In the testing of lithium-ion batteries, ultrasonic waves need to traverse multiple material interfaces, including current collectors, electrodes, and electrolytes. The significant differences in acoustic impedance between these heterogeneous layers easily cause problems such as high-frequency spikes and low-frequency background drift. High-frequency interference often manifests as transient pulse-like fluctuations, originating from electromagnetic coupling or microcracks at the interface; while low-frequency noise may originate from system baseline drift or structural stress relaxation, easily forming long-period fluctuation components. Without effective noise reduction processing of the ultrasonic signals, the accuracy and stability of subsequent diagnostic judgments will be difficult to guarantee.
[0005] Traditional filtering often introduces signal distortions during the purification process, such as blurred waveform edges, peak amplitude shifts, and increased delay errors, leading to irreversible loss of key physical characteristics, especially in material structures with multiple interfaces and multiple reflection paths (such as power batteries).
[0006] To overcome the key problems of existing ultrasonic signal denoising methods, such as strong parameter dependence, poor adaptability to non-stationary signals, and difficulty in balancing denoising and structural feature preservation, this application proposes an adaptive denoising method for EMD decomposition ultrasonic signals based on multi-index fusion evaluation. Summary of the Invention
[0007] To overcome the problems existing in related technologies, this application proposes an adaptive denoising method for EMD decomposition ultrasound signals based on multi-index evaluation, comprising: Empirical mode decomposition (EMD) is performed on the first ultrasonic signal to obtain several IMF components; Feature extraction is performed on several IMF components to obtain signal component features, which include energy proportion, correlation coefficient, Shannon entropy, spectral entropy, and reconstruction error. The IMF components are scored based on the characteristics of the signal components. The IMF components whose scores are greater than or equal to a preset score threshold are superimposed to obtain a second ultrasound signal.
[0008] Preferably, scoring several IMF components based on the characteristics of the signal components specifically includes: Construct a signal component feature matrix for each IMF component using the aforementioned signal component features; The signal component feature matrices of each IMF component are clustered using a clustering algorithm; The weight coefficient of each dimension is dynamically configured based on the projection distance between the cluster centers of each dimension in the signal component feature matrix. The score of each IMF component is calculated using the weight coefficient of each dimension and the characteristics of the signal components.
[0009] Preferably, the step of calculating the score of each IMF component using the weight coefficient of each dimension and the signal component features specifically includes: Substituting the signal component characteristics and the weighting coefficients into the scoring function, the score for each IMF component is calculated; the scoring function is:
[0010] in, For the normalized first The energy percentage of each IMF component For normalized energy percentage weights, For the normalized first The correlation coefficient of each IMF component For the correlation coefficient weight, For the normalized first Shannon entropy of IMF components, For Shannon entropy weights, For the normalized first Spectral entropy of each IMF component For spectral entropy weights, No. Reconstruction error of each IMF component To reconstruct the error weights, For the first The score of each IMF component.
[0011] Preferably, the step of constructing the signal component feature matrix of each IMF component using the signal component features specifically includes: The signal component features are normalized. The signal component feature matrix is constructed by normalizing the signal component features of each IMF component in five dimensions: energy proportion, correlation coefficient, Shannon entropy, spectral entropy, and reconstruction error.
[0012] Preferably, the weight coefficient of each dimension is dynamically configured based on the projection distance between the cluster centers of each dimension in the signal component feature matrix, specifically including: The feature matrix of the current signal components is classified using the K-means clustering algorithm; The weight coefficients are configured based on the projection distance between the cluster centers for each dimension.
[0013] Preferably, the formula for calculating the energy proportion in the signal component characteristics is:
[0014] in, For the first The energy percentage of each IMF component Indicates the first The IMF component in the first The amplitude at each time point For signal length, This represents the total number of IMF components.
[0015] Preferably, the formula for calculating the correlation coefficient in the signal component characteristics is as follows:
[0016] in, For the first The correlation coefficient of each IMF component Indicates that the first ultrasonic signal is at the... The amplitude at each moment, For the first The mean of each IMF component, The mean value of the first ultrasonic signal.
[0017] Preferably, the formula for calculating the Shannon entropy in the signal component features is:
[0018]
[0019] in, For the first Shannon entropy of IMF components, Indicates the first The energy distribution probability of each IMF component Indicates the first The IMF component in the first Energy at a given moment Indicates all IMF components in the th... The total energy at each moment.
[0020] Preferably, the formula for calculating the spectral entropy in the signal component characteristics is:
[0021]
[0022] in, For the first Spectral entropy of each IMF component For the first The spectrum of each IMF component This represents the normalized frequency distribution.
[0023] Preferably, the formula for calculating the reconstruction error in the signal component features is:
[0024] in, For the first Reconstruction error of each IMF component Indicates the removal of the first Reconstructed signal of IMF components.
[0025] The technical solution provided in this application may include the following beneficial effects: This application performs mode decomposition and feature extraction on the first ultrasonic signal, and introduces five types of signal features, namely energy ratio, correlation coefficient, Shannon entropy, spectral entropy and reconstruction error, to score each IMF component. The IMF components with scores higher than the threshold are retained for signal reconstruction. The resulting second ultrasonic signal effectively suppresses high-frequency clutter and low-frequency drift, while retaining the key feature information required for SOC estimation and structural health assessment.
[0026] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and do not limit this application. Attached Figure Description
[0027] The above and other objects, features and advantages of this application will become more apparent from the more detailed description of exemplary embodiments thereof in conjunction with the accompanying drawings, wherein the same reference numerals generally represent the same components in the exemplary embodiments thereof.
[0028] Figure 1 A schematic flowchart illustrating the adaptive denoising method for EMD decomposed ultrasonic signals provided in this application embodiment; Figure 2 To pass Figure 1 A schematic diagram of the IMF components obtained by the method shown; Figure 3 The waveforms of the first and second ultrasonic signals shown in the embodiments of this application are as follows; Figure 4 A pseudo-color image reconstructed from the first ultrasound signal; Figure 5 This is a pseudo-color image reconstructed using the second ultrasound signal. Detailed Implementation
[0029] Preferred embodiments of the present application will now be described in more detail with reference to the accompanying drawings. While preferred embodiments of the present application are shown in the drawings, it should be understood that the present application may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to make the present application more thorough and complete, and to fully convey the scope of the present application to those skilled in the art.
[0030] The terminology used in this application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. The singular forms “a,” “the,” and “the” used in this application and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used herein refers to and includes any or all possible combinations of one or more of the associated listed items.
[0031] It should be understood that although the terms "first," "second," "third," etc., may be used in this application to describe various information, this information should not be limited to these terms. These terms are only used to distinguish information of the same type from one another. For example, without departing from the scope of this application, first information may also be referred to as second information, and similarly, second information may also be referred to as first information. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, "multiple" means two or more, unless otherwise explicitly specified.
[0032] Figure 1 This is a flowchart illustrating the adaptive denoising method for EMD decomposed ultrasonic signals provided in an embodiment of this application. The method includes steps 101 to 104, which are described below in conjunction with... Figure 1 The steps of the method described in this application are explained.
[0033] An exemplary application scenario: The method described in this application can be applied to an ultrasonic monitoring system during the charging process of a lithium-ion battery. Existing ultrasonic detection methods utilize coupling devices and high-frequency transducers to acquire transmission ultrasonic signals during lithium-ion battery charging, obtaining raw time-domain waveform signals containing multi-interface echoes and scattered clutter. These signals exhibit significant non-stationarity and spike-like abrupt changes. The method described in this application can adaptively denoise the raw acoustic echo signal to enhance the accuracy of subsequent feature extraction and state recognition.
[0034] Step 101: Perform empirical mode decomposition on the first ultrasonic signal to obtain several IMF components.
[0035] The first ultrasound signal acquired By applying the empirical mode decomposition method, without any prior basis function setting, the signal is gradually decomposed into several IMF components.
[0036] The format is as follows:
[0037] in, The number of IMF components obtained from the decomposition. Let i be the i-th mode function. like Figure 2 As shown, after performing empirical mode decomposition on the first ultrasonic signal, a total of 10 IMF components were obtained.
[0038] Step 102: Extract features from several IMF components to obtain signal component features.
[0039] Specifically, the signal component characteristics include energy percentage, correlation coefficient, Shannon entropy, spectral entropy, and reconstruction error.
[0040] The calculation methods and applications of the signal component characteristics described below are explained.
[0041] (1) The formula for calculating the energy percentage is:
[0042] in, For the first The energy percentage of each IMF component Indicates the first The IMF component in the first The amplitude at each time point For signal length, This represents the total number of IMF components.
[0043] Energy percentage is used to measure the proportion of a particular IMF component in the total energy of the overall signal. In ultrasonic testing, the main reflection band usually has significant energy, so an IMF with a high energy percentage may contain valid echo information. However, some high-frequency strong interferences (such as electromagnetic noise or transient pulses) also exhibit localized high energy. Therefore, relying solely on energy indicators can easily lead to mistaking high-intensity noise for valid signals; it is necessary to combine other morphological or informational indicators for comprehensive judgment.
[0044] (2) The formula for calculating the correlation coefficient in the signal component characteristics is:
[0045] in, For the first The correlation coefficient of each IMF component Indicates that the first ultrasonic signal is at the... The amplitude at each moment, For the first The mean of each IMF component, The mean value of the first ultrasonic signal.
[0046] This application uses the Pearson correlation coefficient to calculate the correlation coefficient, which is used to evaluate the temporal morphological similarity between a certain IMF and the original signal. The value range of this index is [-1, 1]. The closer the value is to 1 or -1, the higher the structural similarity and feature fidelity between the IMF and the original signal waveform. High correlation indicates that the component makes a positive contribution to the reconstruction of the main waveform of the original signal, and is usually the main carrier of the main reflection or structural echo. If the correlation is close to zero, it means that the IMF is almost unrelated to the original signal and may belong to background noise or invalid components. Therefore, components with high correlation coefficients in IMFs should be retained first to support feature extraction.
[0047] (3) The formula for calculating the Shannon entropy in the signal component features is:
[0048]
[0049] in, For the first Shannon entropy of IMF components, Indicates the first The energy distribution probability of each IMF component Indicates the first The IMF component in the first Energy at a given moment Indicates all IMF components in the th... The total energy at each moment.
[0050] Shannon entropy reflects the statistical uncertainty and information complexity of a signal. In ultrasound signals, clutter and noise components typically exhibit high randomness and uncertainty, resulting in high Shannon entropy values. In contrast, true physical echo signals (such as primary reflections or interface echoes) often exhibit regular, concentrated waveform structures in the time domain, with lower entropy values. Therefore, the orderliness of IMFs can be screened based on entropy values, discarding high-entropy components to purify the signal.
[0051] (4) The formula for calculating the spectral entropy in the signal component characteristics is:
[0052]
[0053] in, For the first Spectral entropy of each IMF component For the first The spectrum of each IMF component This represents the normalized frequency distribution.
[0054] Spectral entropy is used to assess the dispersion of a signal's energy distribution in the frequency domain. In structural ultrasonic testing, effective reflected echoes often have a concentrated dominant frequency and clear spectral characteristics, resulting in low spectral entropy. In contrast, random noise or multipath interference exhibits dispersed frequency distribution and energy diffusion, leading to a significant increase in spectral entropy. Therefore, low spectral entropy components are more likely to contain true echo structures, while high spectral entropy IMFs often represent complex clutter and should be preferentially eliminated.
[0055] (5) The formula for calculating the reconstruction error in the signal component characteristics is:
[0056] in, For the first Reconstruction error of each IMF component Indicates the removal of the first Reconstructed signal of IMF components.
[0057] Reconstruction error is used to assess the impact of removing an IMF on the reconstruction quality of the original signal. If the removal of an IMF has minimal impact on the original signal (i.e., extremely low reconstruction error), it indicates that the IMF does not contain critical information and is highly disposable. Conversely, if the error increases significantly, the IMF is an indispensable and important component of the signal. In practical applications, reconstruction error is the most intuitive quantitative indicator reflecting the value of an IMF, and it is particularly suitable for identifying the boundaries between retaining and discarding edge components.
[0058] The signal component characteristics of the 10 IMF components were calculated using the above calculation method, and the multi-index evaluation results are shown in Table 1.
[0059] Table 1: Multi-indicator evaluation results
[0060] Furthermore, after calculating the signal classification characteristics of each IMF component, in order to unify the evaluation scale, all indicators are first linearly normalized to obtain standardized feature values.
[0061] Let the first The five original characteristics of an IMF are: , , , , These correspond to energy percentage, correlation coefficient, Shannon entropy, spectral entropy, and reconstruction error, respectively.
[0062]
[0063]
[0064]
[0065]
[0066] Considering the significant differences in the dimensions, magnitudes, and directions of the various indicators, directly judging them using thresholds or simple logic would have significant limitations, easily leading to the accidental deletion of valid information or the retention of false signals. Therefore, the method shown in the embodiments of this application proposes a scoring mechanism based on feature normalization and adaptive weighted fusion in step 103, which is used to comprehensively determine and prioritize each IMF component.
[0067] Step 103: Score several IMF components based on the characteristics of the signal components.
[0068] Furthermore, step 103 specifically includes: 201. Construct a signal component feature matrix for each IMF component based on the aforementioned signal component features; Specifically, the normalized features of each IMF component across five dimensions—energy percentage, correlation coefficient, Shannon entropy, spectral entropy, and reconstruction error—are used to construct a five-dimensional vector space, and a feature matrix is built accordingly. .
[0069] 202. Cluster the signal component feature matrices of each IMF component using a clustering algorithm.
[0070] In this embodiment of the application, after performing K-means clustering on the current signal component feature matrix, two cluster centers will be obtained with values in different feature dimensions.
[0071] 203. Based on the projection distance between the cluster centers of each dimension in the signal component feature matrix, dynamically configure the weight coefficient of each dimension.
[0072] In step 203, assuming that when comparing two cluster centers, the difference in the "energy percentage" dimension is 0.7, while the difference in the "correlation coefficient" dimension is only 0.2, then it can be considered that the "energy percentage" feature plays a greater role in distinguishing different clusters, and therefore its weight will be higher than that of the "correlation coefficient". In this way, the weight allocation can be dynamically adjusted according to the magnitude of the feature's discriminative power.
[0073] It is understandable that the weight coefficient of a certain feature dimension is related to the projected distance of that feature dimension between cluster centers, including but not limited to linear and non-linear correlations.
[0074] In battery structure detection tasks, principal components often exhibit clustering characteristics in the directions of correlation and reconstruction error, while noise-type IMFs show a more significant distribution in the direction of entropy-type indicators. The differences between cluster centers can reflect the discriminative ability of each dimension indicator for the current data.
[0075] This application employs unsupervised clustering (K-means) to analyze all IMF components in a five-dimensional feature space, automatically classifying them into signal principal component classes and noise classes. Based on this, the contribution of each feature to the discrimination task is evaluated, thereby achieving dynamic estimation and adaptive updating of weight parameters, avoiding the limitations of fixed weights in traditional methods. This method effectively improves its applicability and stability under different battery structures and operating conditions.
[0076] 204. Normalize the weight coefficients of each dimension, and calculate the score of each IMF component using the weight coefficients of each dimension and the signal component characteristics.
[0077] In step 204, the signal component features and the weighting coefficients are substituted into the scoring function to calculate the score for each IMF component; the scoring function is:
[0078] in, For the normalized first The energy percentage of each IMF component For normalized energy percentage weights, For the normalized first The correlation coefficient of each IMF component For the correlation coefficient weight, For the normalized first Shannon entropy of IMF components, For Shannon entropy weights, For the normalized first Spectral entropy of each IMF component For spectral entropy weights, No. Reconstruction error of each IMF component To reconstruct the error weights, For the first The score of each IMF component.
[0079] Step 104: Superimpose the IMF components whose scores are greater than or equal to a preset score threshold to obtain a second ultrasound signal.
[0080] Based on the above five indicators, the physical meaning and fidelity value of each IMF component are comprehensively evaluated. IMFs with high high-frequency entropy and low correlation (such as IMF1) are spike interference or high-frequency noise and should be eliminated. IMFs with medium frequency, high energy and high correlation (such as IMF2 and IMF3) usually carry the reflection information of the main structure and should be retained. IMFs with low frequency, low energy and high entropy (such as IMF7 and later) are mostly background drift or trend terms and can be discarded.
[0081] exist Figure 2 Of the 10 IMF components shown, IMF2 is the dominant component, accounting for 76.7% of the energy. Its correlation coefficient with the original signal reaches 0.8841, significantly higher than other components. Its Shannon entropy and spectral entropy are 4.72 and 4.59 respectively, both at a low to medium level, indicating strong structural regularity and concentrated frequency domain distribution, making it the main carrier of the primary reflection signal. IMF3, on the other hand, may carry some secondary reflections or structural details. Its energy share is 14.4%, and its correlation coefficient is 0.5491. Although slightly lower than IMF2, it still has value for preservation, and its entropy and spectral entropy do not exhibit typical characteristics of high-frequency noise.
[0082] IMF1 has a low energy percentage (0.74%), but its Shannon entropy (8.65) and spectral entropy (7.97) are the highest in the group. Combined with its sharp abrupt changes in the time domain, it is typically a high-frequency electromagnetic interference or system impulse noise and should be eliminated. IMF4 to IMF10 all show low energy percentages, near-zero correlation, high entropy values, and extremely small reconstruction errors, indicating that these components have very limited role in signal fidelity reconstruction and mainly represent low-frequency background or residual noise without structural significance.
[0083] In step 104, the selected retained IMF components are re-superimposed based on the scores to form the purified and denoised signal:
[0084] in, To preserve the set of indices for the components, This is the second ultrasonic signal.
[0085] In such Figure 2Of the 10 IMF components shown, IMF2 has a high overall score of 0.9734, significantly higher than the other components, verifying its optimal signal fidelity characteristics in the multidimensional index space. This component not only has the highest energy proportion and correlation coefficient (76.7% and 0.8841 respectively), but its entropy level is also in the low-to-medium range, indicating a good balance between preserving structural information and suppressing interference. While IMF3's overall score of 0.396 is significantly lower than IMF2, its moderate energy proportion (14.4%) and moderate correlation still suggest it has some preservation value. In contrast, the scores of the remaining IMFs are generally below 0.2, especially IMF4 to IMF10, which exhibit low energy, low correlation, high entropy, and minimal reconstruction error. Their weight accumulation effect in the fusion scoring function is extremely weak, and they can be considered redundant components with low signal value, thus being eliminated entirely.
[0086] like Figure 3 As shown, the reconstructed second ultrasonic signal significantly suppresses high-frequency pulses and background fluctuations in the time domain, significantly improves the signal-to-noise ratio, and has a clear peak structure, which is beneficial for subsequent feature extraction, such as high-precision calculation of acoustic indicators like dominant frequency, envelope area, and time of flight.
[0087] To verify the effectiveness of the proposed denoising method in spatially distributed structure identification, an ultrasonic imaging experiment of a lithium-ion battery was conducted. By controlling the transducer to perform a gridded scan within a two-dimensional region on the battery surface, the ultrasonic echo signal at each location was acquired and processed point by point, and the amplitude of the denoised signal was mapped to the corresponding spatial coordinates to finally construct a C-mode ultrasonic image.
[0088] in, Figure 4 A pseudo-color image reconstructed from the original ultrasound data. Figure 5 The image shows the result after denoising and feature enhancement using the method of this invention. A comparison clearly shows that the original image suffers from severe overlap between strong reflective areas and background interference, making it difficult to discern key details; while the processed image shows a significant improvement in overall contrast and structural clarity. Especially in the area marked by the black box in the image, multiple hidden defects or microscopic reflective anomalies can be observed, clearly revealing microstructural features.
[0089] The solution of this application has been described in detail above with reference to the accompanying drawings. In the above embodiments, the descriptions of each embodiment have different emphases; parts not described in detail in a certain embodiment can be referred to in the relevant descriptions of other embodiments. Those skilled in the art should also understand that the actions and modules involved in the specification are not necessarily essential to this application. Furthermore, it is understood that the steps in the method of this application embodiment can be adjusted, combined, and deleted according to actual needs, and the modules in the device of this application embodiment can be combined, divided, and deleted according to actual needs.
[0090] Furthermore, the method according to this application can also be implemented as a computer program or computer program product, which includes computer program code instructions for performing some or all of the steps in the method described above.
[0091] Alternatively, this application may be implemented as a non-transitory machine-readable storage medium (or computer-readable storage medium, or machine-readable storage medium) storing executable code (or computer program, or computer instruction code) thereon, which, when executed by a processor of an electronic device (or electronic device, server, etc.), causes the processor to perform part or all of the steps of the methods described above according to this application.
[0092] Those skilled in the art will also understand that the various exemplary logic blocks, modules, circuits, and algorithm steps described in connection with the present application can be implemented as electronic hardware, computer software, or a combination of both.
[0093] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems and methods according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than those marked in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.
[0094] The various embodiments of this application have been described above. These descriptions are exemplary and not exhaustive, nor are they limited to the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments. The terminology used herein is chosen to best explain the principles, practical application, or improvement of the technology in the market, or to enable others skilled in the art to understand the embodiments disclosed herein.
Claims
1. An adaptive denoising method for EMD decomposed ultrasonic signals based on multi-index evaluation, characterized in that, include: Empirical mode decomposition (EMD) is performed on the first ultrasonic signal to obtain several IMF components; Feature extraction is performed on several IMF components to obtain signal component features, which include energy proportion, correlation coefficient, Shannon entropy, spectral entropy, and reconstruction error. The IMF components are scored based on the characteristics of the signal components. The IMF components whose scores are greater than or equal to a preset score threshold are superimposed to obtain a second ultrasound signal.
2. The method as described in claim 1, characterized in that, Scoring of several IMF components based on the characteristics of the signal components specifically includes: Construct a signal component feature matrix for each IMF component using the aforementioned signal component features; The signal component feature matrices of each IMF component are clustered using a clustering algorithm; The weight coefficient of each dimension is dynamically configured based on the projection distance between the cluster centers of each dimension in the signal component feature matrix. The score of each IMF component is calculated using the weight coefficient of each dimension and the characteristics of the signal components.
3. The method as described in claim 2, characterized in that, The calculation of the score for each IMF component using the weight coefficient of each dimension and the signal component features specifically includes: Substituting the signal component characteristics and the weighting coefficients into the scoring function, the score for each IMF component is calculated; the scoring function is: in, For the normalized first The energy percentage of each IMF component For normalized energy percentage weights, For the normalized first The correlation coefficient of each IMF component For the correlation coefficient weight, For the normalized first Shannon entropy of IMF components, For Shannon entropy weights, For the normalized first Spectral entropy of each IMF component For spectral entropy weights, No. Reconstruction error of each IMF component , For the first The score of each IMF component.
4. The method as described in claim 3, characterized in that, The step of constructing the signal component feature matrix of each IMF component using the signal component features specifically includes: The signal component features are normalized. The signal component feature matrix is constructed by normalizing the signal component features of each IMF component in five dimensions: energy proportion, correlation coefficient, Shannon entropy, spectral entropy, and reconstruction error.
5. The method as described in claim 2, characterized in that, Based on the projection distance between the cluster centers of each dimension in the signal component feature matrix, the weight coefficient of each dimension is dynamically configured, specifically including: The feature matrix of the current signal components is classified using the K-means clustering algorithm; The weight coefficients are configured based on the projection distance between the cluster centers for each dimension.
6. The method as described in claim 1, characterized in that, The formula for calculating the energy proportion in the signal component characteristics is as follows: in, For the first The energy percentage of each IMF component Indicates the first The IMF component in the first The amplitude at each time point For signal length, This represents the total number of IMF components.
7. The method as described in claim 6, characterized in that, The formula for calculating the correlation coefficient in the signal component characteristics is as follows: in, For the first The correlation coefficient of each IMF component Indicates that the first ultrasonic signal is at the... The amplitude at each moment, For the first The mean of each IMF component, The mean value of the first ultrasonic signal.
8. The method as described in claim 7, characterized in that, The formula for calculating the Shannon entropy in the signal component features is as follows: in, For the first Shannon entropy of IMF components, Indicates the first The energy distribution probability of each IMF component Indicates the first The IMF component in the first Energy at a given moment Indicates all IMF components in the th... The total energy at each moment.
9. The method as described in claim 8, characterized in that, The formula for calculating the spectral entropy in the signal component characteristics is as follows: in, For the first Spectral entropy of each IMF component For the first The spectrum of each IMF component This represents the normalized frequency distribution.
10. The method as described in claim 9, characterized in that, The formula for calculating the reconstruction error in the signal component features is as follows: in, For the first Reconstruction error of each IMF component Indicates the removal of the first Reconstructed signal of IMF components.